Following the Insulation Line Through Every Junction

The method is the record's own. Take a section, draw the insulation as a continuous line, and note every point at which it will not draw. The record lists what forces those points: a beam, a column, a slab edge, a lintel, a wall plate or a sill that crosses or touches the line because the structure requires it there, together with fixings, ties and support brackets that pass through it to hold cladding, rails and leaves back to the structure.

It then adds a category that has nothing to do with structure. Geometry acts as an interruption in its own right, without any material changing, which is why the record says bridging has to be read off a section rather than off a specification. That is also why the exercise is worth doing on plan as well: the air barrier record asks for the same continuity test in both directions.

The record names the floor perimeter as the junction between floor insulation and wall insulation, sitting at the point where the slab, the barrier, the wall base and often a threshold all meet, built early, buried immediately and impossible to revisit after completion. The damp barrier record adds that this line is built at the busiest and least tidy stage of a project, under weather, alongside groundworks and often before a drawing package is complete.

At a doorway the same line has to be made in an even shorter length. The threshold record describes the floor edge, the wall base and the frame each interrupting the thermal layer within the width of one opening, and notes that cutting the layer to make room for a channel or a fixing puts the coldest internal surface exactly where the most water is.

The published interface record for the slab edge makes a distinction worth holding onto. A dense structural element reaching the perimeter is a thermal path, and unlike a bracket it is not a point but a continuous line around every storey, which makes it one of the largest single contributors to a building's linear heat loss.

The wall records describe what that means in practice. In a light steel infill wall the layers stop at each floor edge, so the weather layer, the air line and the insulation all have to be carried across a band belonging to the primary frame rather than to the wall. In a masonry infill wall the frame members the wall does not occupy are conductive elements running through the envelope, and how the insulation crosses them decides whether they sit inside or outside the thermal line.

The window record describes an opening perimeter as the place where the wall build-up is locally thinnest and where the frame, fixings and structure all cross the insulation line. The bridging record adds the decision that governs it: where a frame sits in the depth of a wall determines whether the insulation can be carried onto it or has to stop short, and the same window in the same opening produces a different junction depending on that position.

The closers, reveals and continuity pieces that close the line there are described as sitting in the tightest parts of the construction and being installed by whichever trade reaches them, so a detail that cannot physically be built will be improvised on site. That is a buildability point with a thermal consequence.

The published interface record is explicit that these are not incidental. Insulation applied to a backing wall is continuous until something has to pass through it, and a supported facade needs its brackets to reach the structure, so every bracket is a deliberate hole carrying a metal component across the layer. As walls become better insulated, these point crossings become a proportionally larger share of the loss.

It also removes the obvious response. The number of brackets follows from the loads and is not a free variable, which is why the record describes the answer as a designed break within the connection rather than simply fewer brackets. The bridging record adds that fixings and closers work against each other, and that an insulation line continuous on the architect's section is routinely crossed by the cladding setting-out and thinned by the window schedule.

At the eaves the line has to close in a space that narrows exactly where the two layers need to overlap, and the junction record describes the requirement as genuinely contradictory where ventilation also enters there. At a parapet the interface record describes a length of construction with outside air on its inner face, its outer face and its top, and says that if the two layers stop where they arrive instead of joining round it, the wall acts as a fin along a line running the whole perimeter of the roof.

On the roof itself the line has to turn corners. The rafter-line record lists the wall head, the ridge, the verge and each dormer, and says that where those turns are incomplete the line is broken. The warm deck record lists the perimeter, the upstands and the penetrations, and says the same.

Some interruptions cannot be closed at all, and the records say so rather than pretending otherwise. Internal wall insulation leaves every intermediate floor, party wall and internal partition crossing the line, and the record states that those junctions cannot all be treated, so the assessment has to address returns, flanking construction and the surfaces that will now run colder. Carrying insulation along those surfaces lengthens the path but never removes it, and each return consumes reveal depth or room dimension that may not be available.

External insulation moves rather than removes the problem: the record says carrying insulation across the outside of a structure removes many junction interruptions at once and replaces them with a smaller set at openings, at the base, at the eaves and wherever brackets pass through. It also names a stopping point that is not physical at all, where the wrap stops at a party boundary or an attached structure, and observes that who owns that edge is as much a legal question as a technical one.

One class of interruption cannot be closed by substituting a better material, because the crossing element has to carry load across the line. The thermal break record describes that as the boundary between the two systems, and says the interruption cannot simply be filled with insulation because the filling would have to carry the load.

It also names the failure that looks like success. A break designed to interrupt conduction while the surrounding insulation stops short simply relocates the path rather than removing it, so the insulation adjoining the break has to close onto it rather than stop at its face. Everything structural about such a connection is determined by a qualified structural engineer.